Are you seeing more cancelled pledges on Kickstarter?

Why are people cancelling pledges? Do we have a sense of who they are? What do we really know about how backers make decisions on crowdfunding platforms?

Hope you’re all doing well and enjoying your weekend.

I’ve been busy working on Kickstarter fulfilment for my recent campaign, while also picking through the bones of the data to get a better sense of where things could be improved for next time.

One thing I wanted to talk about today is cancelled pledges.

Photo by Markus Winkler on Unsplash

I’ve noticed an uptick in the number of adjusted and cancelled pledges during recent campaigns. Adjusted pledges to me indicate people are being more selective about project backing (almost always the adjustment is from a higher to lower reward tier).

But the cancelled pledges are interesting, because they come from people who had backed large numbers of projects, or were otherwise identified as super backers.

Why are they cancelling their pledges?

Maybe they just don’t like the project? But if that’s the case, then why did they back it in the first place?

I don’t back projects unless I want the product (be it comic, or book, or something else), but I have to accept that other people don’t think this way.

However, these are super backers who are cancelling their pledges, so is there something specific to this class of backers that makes them more likely to behave this way?

Photo by Jon Tyson on Unsplash

In any event, I was curious about understanding this further, and went searching for some research. While there are lots of papers and reviews about purchasing behaviour for marketing, retail, and e-commerce, there wasn’t much for crowdfunding.

While crowdfunding can be thought of as an online sales platform, it’s not the same as retail for many reasons, but one that is recurrent in the literature is information asymmetry between creator and backers. The backer has to make a judgment based on the story provided by the creator (and perhaps other inferred social signals) about whether i) the product will exist in the described form, and ii) it will reach them in the timelines provided.

I found a very interesting paper on this topic, Incomplete Decisions on Reward-Based Crowdfunding Platforms: Exploring Motivations from Temporal and Social Perspectives, which I’ll try to break down in this essay.

The authors rightly point out that assumptions of rational consumer decision-making may not be true in reward-based crowdfunding, which prompted them to be careful in how they approached their study. They also noted that crowdfunding creates time-sensitivity for backers, which can influence their decision-making, making it more likely that they prefer to make spontaneous funding decisions rather than deliberate ones that utilize all the available information - which they refer to as incomplete decisions.

As an aside: I nodded as I read this in their work. It’s actually quite rare for me to read all the information provided about a given project. Once I get a sense or feel for what it is, I either back it or I don’t. I almost never watch the campaign video (and am aware that few people watch mine). Again, this is just my perspective, but it aligns exactly with their initial background.

To probe incomplete decisions in the crowdfunding context, the study uses the Stimulus Organism Response framework, which is a widely used framework in retail behaviour modelling. The framework uses a three-step process: introducing stimuli, processing these stimuli, and finally behavioural response. This model is the theoretical underpinning for the rest of their work.

The paper focuses on testing three hypotheses, which are based on key factors affecting backer decision-making.

• The earlier the campaign phase, the less probable it is to encounter incomplete decision-making.

• The greater the current pledge level, the likelihood of encountering incomplete decisions is lower.

• There exists a positive interaction effect between campaign phase and current pledged amount.

The first hypothesis essentially argues that in the early days of a campaign, initial backers are more risk tolerant, and more likely to be receptive to all the information provided about the campaign. Whereas, as time runs out, backers decide to back (or not) based on this factor rather than assessing all the information.

The second hypothesis argues that projects with low levels of funding are higher risk for backers (compared to those with much higher funding levels). This risk - they argue - should lead backers to make more informed decisions using all the available information. However, it is tempered by other factors, such as altruistic motivators, or even contrarian motivators for projects that are over-funded.

Finally, the third hypothesis is basically stating that a mix of these first two is actually more realistic.

For their experiment, they used a 2X2 design to model four states: early (first 3 days) vs late (last 3 days of a campaign); and pledge level (300% funded vs. less than 30%).

They identified 28 campaigns (with 7 campaigns each associated with one of the given states) on Kickstarter, and then used a survey to send to around 400 potential backers. The survey asked for initial pledge decisions, it was then followed by an information-literacy intervention, and finally a re-assessment of their pledging decision. By careful tracking of the data, the study hoped to discover if backers changed their minds, and which pieces of information influenced their decisions.

Findings from the study

From their analysis, almost 51% of backers changed their minds after the intervention.

Logit regression analysis also confirms hypothesis one and two, in models where those are the variables of interest. For example, the model showed an almost 300% increased likelihood of backers changing their decision during the last three days of the campaign, compared to the first three days. Similarly, their second model shows an 84% reduced likelihood of backers changing their decision when the current pledge level exceeds 300%.

However, it’s the third hypothesis which is more useful to creators, as that is the more realistic scenario.

The regression found a negative and statistically significant coefficient, which indicates that campaign phase decisions are influenced by the current pledge level. However, the surprise was that the control variable had an even larger effect.

• More experienced backers were significantly more likely to change their pledge (for both positive and negative impact), compared to inexperienced backers.

• Older backers were significantly more likely to change their pledge in a negative direction (with reference to their initial decision) compared to younger backers.

For the campaign phase model, backers were more likely to alter their decision (from not backing to backing) when facing tighter time constraints

For the current pledge amount model, backers also exhibit herding behaviour which reduces uncertainty around initial decision-making, which also makes it less likely they will change their minds.

The interaction model suggests backers are more likely to change their initial decisions under heightened time pressure and also when it appears the campaign will be unsuccessful.

Photo by Alexander Schimmeck on Unsplash

What does this mean for creators

The finding that over half of all backers change their minds, is pretty mind-blowing.

My question was answered in this paper, albeit unsatisfactorily: more experienced backers (which would include super backers) are more likely to change their minds than other backers. The authors suggest this is due to more nuanced and calculated decision-making, although they unfortunately provide no further details about the motive for this change.

This is very interesting, as previous articles have explored the importance of herding effects in the early stages of a campaign, where it was important to try and get super backers to back the campaign early.

While this is still good advice for having a good launch, the article here suggests these backers may not actually stick around for the duration of the campaign.

For creators, the trick is then to entice super backers early, and ensure the campaign gets funded early, so if (or when) these super backers cancel their pledges, your campaign doesn’t fall back below the funding threshold.

Of course, another action creators could take is to try and retain super backers for the duration of the campaign through ongoing messaging. Given the seemingly mercurial nature of super backers, it’s not always clear what those messages should be, but perhaps that’s not the point. Using the messaging function, you can continue to be present in their inbox even without sending updates. Maybe that alone would be enough to prevent a cancellation? Who knows. It’s something I intend to test on future campaigns.

A second key insight from the paper is that decisions are made by backers viewing only a select few pieces of information, rather than the full amount of information provided by the creator¹. The authors suggest providing more customized information to backers tailored to interest, age, education, and crowdfunding experience.

This is fascinating and worth exploring. My own experience is that I typically craft my campaign using a template, that speaks to emotional and intellectual levers. However, this does mean the story pages can be quite long, and difficult to parse. In serialized campaigns, it also means backers who are familiar with previous instalments may feel forced to read information they already know.

I suspect this would mean adopting a new template, that breaks the story section down further, providing different information for old and new backers, or those familiar with a genre vs., those new to a genre (for example). Naively, I’m not sure this would work with the standard linear structures in place, but it’s certainly worth considering.

In many respects, it makes me wonder about what is critical information for backers. For exampe, I’ve explored many papers where the inclusion of a video was associated with higher success rates.

I believe this finding is correct, but only in the aggregate.

These studies never explore category influence, which is where I think we would see a change as there is a big difference between visual end-products such as comics, where pages of artwork are often included in the campaign story, and other products such as prototypes, software, events etc, where you need to hear from the creator about their vision. Given this, I feel that - for comic creators - the video could be an information element to skip.

My takeaway is that I think more targeted messaging is critical for engagement, and that may mean subtle shifts to how I approach campaign page design. It’s impossible to be everything to everyone, but I think a more specific approach to information sharing is a good idea. What that looks like in practice though, is still anyone’s guess.

---

That’s it for this week. What tips do you think could make your campaign stickier for super backers? What do you think about targeted story information? Always happy to hear your thoughts.

1. I’m focused on those creators who provide a lot of information about their projects. As we know, some creators (mostly spammers or those naive to the platform) provide little/no information about what they intend to make.